Keywords

Parameter sensitivity, calibration, snow model, multiobjective evolutionary algorithm

Location

Session C1: Compexity, Sensitivity, and Uncertainty Issues in Integrated Environmental Models

Start Date

16-6-2014 9:00 AM

End Date

16-6-2014 10:20 AM

Abstract

Recent projections of environmental change have shown possible variation to temperature and precipitation patterns due to climate change. In snowmelt-dominated watersheds, adapting to such environmental changes requires a detailed understanding of hydrological processes in addition to historical snow cover and streamflow data. Snow models are often incorporated as an additional component of hydrological modeling studies that inform research and operations management. However, previous research and parameter estimation approaches using snow models assume that parameters have a single optimal value and that each parameter is sensitive. This paper demonstrates that an improved understanding of snow model parameter sensitivity can aid in key decisions for water management in these basins. This study uses Sobol's sensitivity analysis method combined with a multi-objective optimization calibration to determine a set of sensitive parameters for a snow model and obtain parameter sets that perform well with respect to multiple calibration objectives. The calibration results provide an ensemble of possible hydrological outcomes that can help management decisions. Data from four Natural Resource Conservation Service SNOTEL sites in Colorado were chosen to demonstrate the approach.

Share

COinS
 
Jun 16th, 9:00 AM Jun 16th, 10:20 AM

Investigating Parameter Sensitivity for Management in Snow-Driven Watersheds

Session C1: Compexity, Sensitivity, and Uncertainty Issues in Integrated Environmental Models

Recent projections of environmental change have shown possible variation to temperature and precipitation patterns due to climate change. In snowmelt-dominated watersheds, adapting to such environmental changes requires a detailed understanding of hydrological processes in addition to historical snow cover and streamflow data. Snow models are often incorporated as an additional component of hydrological modeling studies that inform research and operations management. However, previous research and parameter estimation approaches using snow models assume that parameters have a single optimal value and that each parameter is sensitive. This paper demonstrates that an improved understanding of snow model parameter sensitivity can aid in key decisions for water management in these basins. This study uses Sobol's sensitivity analysis method combined with a multi-objective optimization calibration to determine a set of sensitive parameters for a snow model and obtain parameter sets that perform well with respect to multiple calibration objectives. The calibration results provide an ensemble of possible hydrological outcomes that can help management decisions. Data from four Natural Resource Conservation Service SNOTEL sites in Colorado were chosen to demonstrate the approach.